Ip adapter automatic1111 sdxl

Ip adapter automatic1111 sdxl. IP-Adapter FaceID. bin. Setup ControlNet & IP Adapter Models. Negative prompt: ugly, deformed Dec 17, 2023 · こちらは先日アップデートされた AUTOMATIC1111 Ver1. In ComfyUI on the other hand, you can perform all of these steps in a single click. 5 workflow, where you have IP Adapter in similar style as the Batch Unfold in ComfyUI, with Depth ControlNet. 5 or lower strength, so not great likeness. Feb 10, 2024 · SDXLが高速化&標準で多様な機能が搭載 | イクログ. ControlNet 1 settings. Reuploaded as . Good news everybody - Controlnet support for SDXL in Automatic1111 is finally here! (Now with Pony support?) This collection strives to create a convenient download location of all currently available Controlnet models for SDXL. Yeah, Fooocus is why we don't have an Inpainting CN model after 6-7 months: the guy who makes Fooocus used to make ControlNet and he dumped it to work on Fooocus. (Make sure that your YAML file names and model file names are same, see also YAML files in "stable-diffusion-webui\extensions\sd-webui-controlnet\models". 6 there are plenty of new opportunities for using ControlNets and sister models in A1111,Get the COMFYUI SDXL Advanced co Force use CPU for specified modules, default: [] --listen Launch web server using public IP address, default: False --port PORT Launch web server with given server port, default: 7860 --freeze Disable editing settings --auth AUTH Set access authentication like "user:pwd,user:pwd"" --auth-file AUTH_FILE Set access authentication using file Automatic1111 is the most powerful opensource web user interface for Stable Diffusion. ②WebUIで使用する 2つのモデルを同時に使用する必要があるようです。デフォルトでマルチコントロールネットですね。 順番が決まっているようです。最初にIP-adapterモデル、次にコントロールネットモデルを使用する。 Mar 4, 2024 · Home · lllyasviel/stable-diffusion-webui-forge Wiki (github. safetensors, SDXL face model ip-adapter_sdxl. Also, its ability to blend faces effectively and maintain consistency across various prompts and seeds is quite remarkable (have a look at the image below). (5) Set the Control Weight to 0. This duo works together to detect and refine facial Feb 11, 2024 · I used controlnet inpaint, canny and 3 ip-adapter units each with one style image. Enable a Controlnet. With the prompt: A woman sitting outside of a restaurant in casual dress. safetensors. Jan 22, 2024 · While I haven't tested it thoroughly, it seems like the portrait IP-adapter might be faster than others from the faceid family. You are not restricted to use the facial keypoints of the same person you used in Unit 0. First, open ComfyUI navigate to " Manager " and click " Update All " to update ComfyUI and the nodes. Finally, these tips might help fine-tune your results: If the color of the new attire isn't quite right, consider expanding the mask slightly to capture the surrounding hues. 0 の個人的な設定や、拡張機能の覚書です。 以前の記事に乗せていたのですが、Settingの項目が大幅にリニューアルされまして、同じ設定をしようにも迷ってしまいましたので、改めて書き出しておこうと思います。 Especially if any changes is made by AUTOMATIC1111 which would completely break for forge which forge would have to not merge. It's an ideal solution for artists, designers, and anyone Aug 18, 2023 · ip-adapter_sdxl. 5は「ip-adapter_sd15. 400 supports beyond the Automatic1111 1. ai are marked as fp32/fp16 only to make it possible to upload them both under one version. Wiki Home. pth」、SDXLなら「ip-adapter_xl. 0 with AUTOMATIC1111. X, and SDXL. 手順2：必要なモデルのダウンロードを行い、所定のフォルダに移動する. Drag the image to be inpainted on to the Controlnet image panel. Use TAESD; a VAE that uses drastically less vram at the cost of some quality. 4 for ip adapter and for the prompt I used a very high weight for the "anime" token. Getting consistent character portraits generated by SDXL has been a challenge until now! ComfyUI IPAdapter Plus (dated 30 Dec 2023) now supports both IP-Adapter and IP-Adapter-FaceID (released 4 Jan 2024)! I will be using the models for SDXL only, i. 0. 👍 1 clayne reacted with thumbs up emoji With a ControlNet model, you can provide an additional control image to condition and control Stable Diffusion generation. Install ip-adapter; Enable it in ControlNet; Not works. Hit Generate and now we can see that we have exactly what we want 😄. 5), T2I-Adapter-SDXL still uses the original recipe, driving 2. AUTOMATIC1111 can run SDXL as long as you upgrade to the newest version. 9 のモデルが選択されている. Features included: Support for SDXL1. txt. Sep 8, 2023 · Compared to previous versions of T2I-Adapter (SD-1. Feb 1, 2024 · ip-adapter_instant_id_sdxl control_instant_id_sdxl. You can use it without any code changes. Make sure to adjust the control weights accordingly so that they sum up to 1. Select the SDXL checkpoint that you want to use. Each of them is 1. Sysinfo. Use the same resolution for generation as for the original image. Important: set your "Starting Control Step" to 0. The Original IP-adapter The journey begins with the Original IP Nov 30, 2023 · #a1111 #stablediffusion #fashion #ipadapter #clothing #controlnet #afterdetailer #aiimagegeneration #tutorial #guideThe video talks mainly about uses of IP Mar 7, 2023 · 1. 0 Light impact model Sep 5, 2023 · Kohya氏の「ControlNet-LLLite」モデルを使ったサンプルイラスト. With IP-Adapter model and ControlNet extension of Automatic1111 Web UI you can change images as below with and without prompting 16 Like Comment Mar 10, 2024 · The current update of ControlNet1. safetensors , v1. Aug 3, 2023 · Extremely slow SDXL image generation I am using RTX3070Ti Laptop GPU with 8G of VRAM. Step 0: Get IP-adapter files and get set up. SDXL 0. Many of the new models are related to SDXL, with several models for Stable Diffusion 1. Generative Fill in Photoshop) is really useful in many workflows, but not straight forward with SD. 3-0. Steps to reproduce the problem. 0, which comes with 2 models and a 2-step process: the base model is used to generate noisy latents , which are processed with a refiner model specialized for the Jan 11, 2024 · * One-Click Installation & Usage: Discover the simplicity of using the IP Adapter Face ID with my custom-coded Gradio application. SDXLは基本の画像サイズが1024x1024なので、デフォルトの512x512から変更してください Nov 3, 2023 · The key is that your controlnet_model_guess. Inpaint as usual. 9のモデルが選択されていることを確認してください。. Step 3: Enter ControlNet settings. Jan 7, 2024 · Posted7 Jan 2024. The Foundation: Installing ControlNet on Diverse Platforms :Setting the stage is the integration of ControlNet with the Stable Diffusion GUI by AUTOMATIC1111, a cross-platform software free of charge. 7 to avoid too high weight to interfere with the output. 1 and Different Models in the Web UI - SD 1. 0 version. Select Controlnet Control Type "All" so you can have access to a weird Feb 29, 2024 · A Deep Dive Into ControlNet and SDXL Integration. (2) Set the Control Type to: IP-Adapter. g. 5 ControlNet models – we’re only listing the latest 1. It's a massive quality improvement over previous models, however runs quite slowly on Macs. Step 4: Generate images. The faces look like if I had trained a LORA and used . Step 1. There was this excellent discussion some months ago which uses Auto1111, ControlNet inpaint_only Dec 14, 2023 · ControlNet is a crucial component of Stable Diffusion XL (SDXL) that helps create stable and stunning art. Detected Pickle imports (3) Transfer Clothing Style using Automatic1111 & IP AdapterIP Adepter (ip-adapter-plus_sdxl_vit-h)Background removal extension for A1111 (stable-diffusion-webui Nov 18, 2023 · Contents include:Sample resultsGeneration using LCM LoRA + Tile+Temporal+Soft edge controlnetGeneration without controlnet using LCM LoRA Generation with usi Feb 19, 2024 · Outpainting with SDXL in Forge with Fooocus model, Inpainting with Controlnet Use the setup as above, but do not insert source image into ControlNet, only to img2image inpaint source. So you should be able to do e. Performance comparison Dec 17, 2022 · SDXL v1. #11 opened 6 months ago by MonsterMMORPG. 6B SDXL with a 79M Adapter! T2I-Adapter-SDXL maintains powerful control capabilities while inheriting the high-quality generation of SDXL! Training T2I-Adapter-SDXL with diffusers ip-adapter-plus-face_sdxl_vit-h. 3 GB VRAM via OneTrainer - Both U-NET and Text Encoder 1 is trained - Compared 14 GB config vs slower 10. Trying to find a nice workflow to make the images in 1. safetensors - ip-adapter-plus_sdxl_vit-h. Adetailer Automatic1111 ControlNet Extensions Mastering Image Prompts in Stable Diffusion AI with IP-Adapter Jan 6, 2024 · Embrace the speed and power of SDXL Turbo on Automatic1111. Just like in the previous steps above, Drop down the ControlNet tab: (1) Click Enable. 3 The ControlNet Models. The newly supported model list: May 2, 2024 · Upgrading to IP Adapter V2: 1. There have been a few versions of SD 1. Select the IP-Adapter radio button under Control Type. Apr 18, 2024 · How to use InstantID on AUTOMATIC1111. Our method not only outperforms other methods in terms of image quality, but also produces images that better align with the reference image. safetensors] Failed to load checkpoint, restoring previous Jul 14, 2023 · 推論の実行. How to use InstantID on ComfyUI. 1. Download the IP Adapter ControlNet files here at huggingface. bin/. Feb 28, 2024 · InstantID uses InsightFace to detect and extract a facial embedding from your chosen face, then pairs it with the IP-Adapter to guide the image generation process. IP-adapter models. 0; Simple plug and play (no manual configuration needed) Training can be done; Multiple ControlNet Working; Multiple options like T2I Adapter, Composer, Control LoRA etc. sysinfo-2023-11-02-09-47. Pro Tip: Share your experiences and discoveries in the Automatic1111 community! Connect with other creators, learn new techniques, and Dec 31, 2023 · Make the following changes to the settings: Check the " Enable" box to enable the ControlNet. You can also a custom models. If not, go to the Extensions tab in Automatic1111 and select Install from URL. Jul 26, 2023 · When i select the sdxl model to load, I get this error: Loading weights [31e35c80fc] from D:\stable2\stable-diffusion-webui\models\Stable-diffusion\sd_xl_base_1. 400 is developed for webui beyond 1. This is well suited for SDXL v1. Let’s see how it’s done. Method 1 (Manual): Apr 27, 2024 · Stable Diffusion 1. Stable Diffusion web UI. Feb 19, 2024 · Outpainting with SDXL in Forge with Fooocus model, Inpainting with Controlnet Use the setup as above, but do not insert source image into ControlNet, only to img2image inpaint source. Txt2img. 💡 FooocusControl pursues the out-of-the-box use of software. 5 プロンプト：girl. Although training does seem to work, it is incredibly slow and consumes an excessive amount of memory. You can use multiple IP-adapter face ControlNets. Console logs Jan 29, 2024 · There is a chance that you named the instant id's controlnet model as ip-adapter_instant_id_sdxl, so that the unit0 is complaining not finding an IPAdapter unit before it to read embedding from. e. Good news everybody - Controlnet support for SDXL in Automatic1111 is finally here! This collection strives to create a convenient download location of all currently available Controlnet models for SDXL. 5 vs 2. 8, 2023. Model: "ip-adapter-plus_sd15" (This represents the IP-Adapter model that we downloaded earlier). Free software usually encounters a lot of installation and use of the problem, such as network problems caused by the May 30, 2023 · Currently most functionality in the web UI works correctly on macOS, with the most notable exceptions being CLIP interrogator and training. (3) Set the Preprocessor to: ip-adapter_clip_sdxl. That’s why SDXL Turbo works on AUTOMATIC1111 without official support! Use a low sampling step 1 – 4. It is a more flexible and accurate way to control the image generation process. 9; sd_xl_refiner_0. Oct 6, 2023 · Saved searches Use saved searches to filter your results more quickly Aug 6, 2023 · LEGACY: If you're interested in comparing the models, you can also download the SDXL v0. 5ベースの内容になります。SDXLの場合は都度お知らせします。 Dec 11, 2023 · I tried to finetune the SDXL IP-Adapter but even after only 2K steps, it's already really fried, I'm only able to produce an image okayish with a weight of the trained IP-Adapter between 0. In this article, we will explore the world of Stable Diffusion Automatic 11 11 and its various extensions called Jan 7, 2024 · Posted7 Jan 2024. Follow these directions if you don't have AUTOMATIC1111's WebUI installed yet. Unit 1 Setting. there is now a preprocessor called gaussian blur. 5 for download, below, along with the most recent SDXL models. 5 controlnet, upscale and img2img with SDXL to push details. It effectively acts like an 'instant LoRA' as @huchenlei Nov 30, 2023 · Karras noise schedule: SDXL Turbo with 1 – 10 steps and Euler ancestral sampler. Now, if you don't have the relevant nodes installed and you are getting a missing node error, then there are two ways to install nodes:-. safetensors) Custom Models. 1. Inpaint/Outpaint without text prompt (aka. 7. The old `ip-adapter-faceid_sdxl_lora. Send the generation to the inpaint tab by clicking on the palette icon Prompt. pt from h94 has to be renamed manually after downloading. IP Adapter XL Canny represents a significant advancement in the field of AI-driven image transformation. Dec 17, 2022 · SDXL v1. 3 days ago · Known issues: currently there is no preprocessor for the blur model by kohya-ss , you need to prepare images with an external tool for it to work. 6. 各項目を見る前に、以下の注意点がございます。 基本的にはSD1. Step 2: Enter the txt2img setting. We collaborate with the diffusers team to bring the support of T2I-Adapters for Stable Diffusion XL (SDXL) in diffusers! It achieves impressive results in both performance and efficiency. Jan 23, 2024 · この動画ではStable Diffusion WebUI(AUTOMATIC1111)でIP-Adapter-FaceIDを使って、同じ顔の人物を生成する方法を具体的に説明します。IP-Adapter-FaceIDを使えば Mar 4, 2024 · As files with the extension . Lately, I have thrown them all out in favor of IP-Adapter Controlnets. , ControlNet and T2I-Adapter. Jan 12, 2024 · IP-Adapterのモデルをダウンロード. There are ControlNet models for SD 1. 当ブログにはプロモーション、アフィリエイト広告が含まれます。. Once the ControlNet settings are configured, we are prepared to move on to our AnimateDiff settings. I had a ton of fun playing with it. 5, SD 2. The two versions of the control-loras from Stability. Nov 2, 2023 · lastest version of control net, automatic1111. 0 ControlNet models are compatible with each other. Closed 1 task done. 以下のリンクからSD1. 703 MB LFS support safetensors 5 months ago; ip-adapter_sdxl_vit-h. If you’re looking to create SDXL art or just interested in Feb 29, 2024 · ControlNet IP-Adapter: This technique allows for inserting specific patterns or images onto the new attire. ) Automatic1111 Web UI - PC - Free How to use Stable Diffusion V2. com) Stable diffusion webui forgeのGitHubサイト内にwikiが出来ていて、コントロールネット置き場のまとめがしてあります。 ここを見れば色々場所を探す必要がないということが分かりました。 ただ、ここに取り上げていないコントロールネットのモデルも Jan 27, 2024 · With IP Adapter face models, you can use a face from your input image and copy that face to your generated image. Built on the SDXL framework, this model integrates the IP Adapter and Canny edge preprocessor to offer unparalleled control and guidance in creating context-rich images. Reply reply More replies More replies AlfaidWalid 3 days ago · ControlNetXL (CNXL) - A collection of Controlnet models for SDXL. pth」をダウンロードしてください。 lllyasviel/sd_control_collection at main. This is quite similar to how the IP-Adapter Face ID operates. 手順3：必要な設定を行う A1111 ControlNet now support IP-Adapter FaceID! Not getting good results with FaceID Plus v2 / SD 1. 1 vs Anything V3. For example, if you provide a depth map, the ControlNet model generates an image that’ll preserve the spatial information from the depth map. Here's a quick how-to for SD1. SDXL FaceID Plus v2 is added to the models list. LARGE - these are the original models supplied by the author of ControlNet. IP-Adapter FaceID provides a way to extract only face features from an image and apply it to the generated image. The sd-webui-controlnet 1. ) Oct 11, 2023 · 『IP-Adapter』とは 指定した画像をプロンプトのように扱える技術のこと。 細かいプロンプトの記述をしなくても、画像をアップロードするだけで類似した画像を生成できる。 実際に下記の画像はプロンプト「1girl, dark hair, short hair, glasses」だけで生成している。 顔を似せて生成してくれた Sep. 左上にモデルを選択するプルダウンメニューがあります。. ) Automatic1111 Web UI - PC - Free Easiest Way to Install & Run Stable Diffusion Web UI on PC by Using Open Source Automatic Installer. Dec 19, 2023 · In AUTOMATIC1111, you would have to do all these steps manually. The update to WebUI version 1. Mar 20, 2024 · Control Type: "IP-Adapter". New installation. ip-adapter-plus-face_sdxl_vit-h and IP-Adapter-FaceID-SDXL below. Jan 10, 2024 · Update 2024-01-24. Looks like you can do most similar things in Automatic1111, except you can't have two different IP Adapter sets. SDXL base 0. The best part is if you use 1 step, the Turbo and Karas schedules are the same. All reactions none of the vit-h models can work in Automatic1111 WebUI ip-adapter Jan 28, 2024 · You must set ip-adapter unit right before the ControlNet unit. 1 How to install this in AUTOMATIC1111? 1 #21 opened 4 months ago by Valachio. The ControlNet unit accepts a keypoint map of 5 facial keypoints. Control Weight: 1; The remaining settings can remain in their default state. So, you need to update the Automatic1111, if not yet. Now we have perfect support all available models and preprocessors, including perfect support for T2I style adapter and ControlNet 1. 5 and Stable Diffusion 2. プロンプトは「girl」だけなので、IP-Adapterの効果が現れていることが分かります。 ReferenceよりもIP-Adapterの方が、元画像に近い画像が生成されました。 IP-adapter-plus-face_sdxl is not that good to get similar realistic face but it's really great if you want to change the domain. 9 models: sd_xl_base_0. Here's the guide on running SDXL v1. The speed of image generation is about 10 s/it (10241024 batch size 1), refiner works faster up to 1+ s/it when refining at the same 10241024 resolu Control Nets for SDXL; Learning Generative AI with SDXL; Stable Diffusion and Comfy UI; Advanced Courses on Automatic 11 11; T2i Adapters and IP Adapters; Installation Process; Using Control Nets for Text-to-Image; Introduction. 0 is pre-requisite for harnessing the SDXL model within this How to use IP Adapter Face ID and IP-Adapter-FaceID-PLUS in sd15 and SDXL. That should not be affecting the main branch. As files with the extension . Feb 18, 2024 · 導入方法：IP-Adapterモデルをダウンロードする 「IP-Adapter」のモデルは、「Hugging Face」の公式ページから入手可能です。 「IP-Adapter」をダウンロードした後に、Stable Diffusion WebUIにインストールします。 導入からインストールまでの手順は以下の通りです。 3 days ago · ControlNetXL (CNXL) - A collection of Controlnet models for SDXL. All reactions Mar 16, 2024 · Preprocessor: ip-adapter_clip_sd15; Model: ip-adapter-plus-face_sd15; The control weight should be around 1. Use the same resolution for inpainting as for the original image. This will increase speed and lessen VRAM usage at almost no quality loss. I used a weight of 0. 9; Install/Upgrade AUTOMATIC1111. There are three different type of models available of which one needs to be present for ControlNets to function. Below is the result this is the result image with webui's controlnet Jul 8, 2023 · After doing research and reading through the various threads here and on Reddit mostly, I think the Linux version of Automatic1111 might have an issue with loading the SDXL model due to a memory leak of some kind. He only does the bare minimum now. SDXL v1. SDXLでControlNetを使う方法まとめ. The IP-Adapter is fully compatible with existing controllable tools, e. 1 and 0. safetensors , vit-G SDXL model, requires bigG clip vision encoder Deprecated ip-adapter_sd15_light. 3. The projected face embedding output of IP-Adapter unit will be used as part of input to the next ControlNet unit. Console log errors; What should have happened? Ip-adapter should work correctly. * Cross-Platform Compatibility: Learn how to operate this tool on Dec 19, 2023 · [Bug]: RuntimeError: mat1 and mat2 shapes cannot be multiplied (1x1024 and 1280x8192) when using SDXL IP-Adapter on M2 Mac #2342. safetensors` seems to be better than the new. It plays an important role in the creation of SDXL art by assisting with the installation, VRAM settings, Canny models, Depth models, Recolor models, Blur models, and IP-Adapter. Comparison of InstantID with pre-trained character LoRAs. We release T2I-Adapter-SDXL models for sketch, canny, lineart, openpose, depth-zoe, and depth-mid. Image-guided image-to-image and inpainting can be also achieved by simply replacing text prompt with Now You Can Full Fine Tune / DreamBooth Stable Diffusion XL (SDXL) with only 10. In contrast, InstantID achieves better fidelity and retain good text editability (faces and styles blend better). Stable Diffusion Web UI forgeのインストール方法や使い方について紹介しています。. Select ip-adapter_clip_sd15 as the Preprocessor, and select the IP-Adapter model you downloaded in the earlier step. This is the Stable Diffusion web UI wiki. (4) Set the ControlNet model to: ip-adapter_xl. Let's . ) Feb 28, 2024 · This guide will introduce you to the full range of IP-Adapter models, including the Plus, Face ID, Face ID v2, and Face ID portrait variants, and will provide detailed instructions on how to implement IP-Adapters in the AUTOMATIC1111 and ComfyUI interfaces. Let your imagination run wild, craft breathtaking visuals, and share your creations with the world! Remember, the only limit is your own creativity. Jan 11, 2024 · IP-Adapterを使った結果は、以下のとおりです。 SDXL プロンプト：girl SD1. We release two online demos: and . I also tried ip-adapter image with original sizes and also cropped to 512 size but it didn't make any difference. pth」か「ip-adapter_sd15_plus. Step 1: Select a SDXL model. Make sure ControlNet is installed in your Automatic1111 WebUI. With a ControlNet model, you can provide an additional control image to condition and control Stable Diffusion generation. However, the results seems quite different. 45 GB large and can be found here. 3 GB Config - More Info In Comments Dec 14, 2023 · Model weights: Use sdxl-vae-fp16-fix; a VAE that will not need to run in fp32. I think these 2 file names are mixed ip-adapter-plus-face_sdxl_vit-h. py file can not recognize your safetensor files, some launchers from bilibili have already included the codes that @xiaohu2015 mentioned, but if you're using cloud services like autodl, you need to modify codes yourself, as those dockers are using the official controlnet scripts . Either the Turbo or Karras noise schedules are fine. Mar 4, 2024 · ControlNetXL (CNXL) - A collection of Controlnet models for SDXL. 0 (download link: sd_xl_base_1. 0 has just been released. 1 versions for SD 1. 手順1：Stable Diffusion web UIとControlNet拡張機能をアップデートする. ip-adapter-face-id-automatic1111 ip-adapter-faceid-plus ip-adapter-faceid-plusv2; Oct 24, 2023 · In addition to controlnet, FooocusControl plans to continue to integrate ip-adapter and other models to further provide users with more control methods. Or you can have the single image IP Adapter without the Batch Unfold. Tips for getting good images. 5. Dec 5, 2023 · I'm using IPAdapter here, and I've chosen the model ip-adapter-plus_sd15 and set the weight to 0. 1 Shuffle. 4/1. Send the generation to the inpaint tab by clicking on the palette icon It can be seen that both PhotoMaker and IP-Adapter-FaceID achieves good fidelity, but there is obvious degradation of text control capabilities. What browsers do you use to access the UI ? Google Chrome. ControlNet 0 settings. In addition, I have prepared the same number of OpenPose skeleton diagrams as the uploaded movie and placed them in the /output/openpose folder for this ControlNet to read, Sep 21, 2023 · T2I-Adapter; IP-Adapter; 結構多いです。これを使いこなせている人はすごいですね。次は各項目の解説をしていきます。 各項目を見る前に. Sep 8, 2023 · With the arrival of Automatic1111 1. Preprocessor: "ip-adapter_clip_sd15". I'm talking about 100% denoising strength inpaint where you just have to select an area and push a button. I know that there is some non-oficial SDXL inpaint models but, for instance, Fooocus has his own inpaint model and works pretty well. the SD 1. The secret sauce of InstantID is its combination with ControlNet. Standard SDXL inpainting in img2img works the same way as with SD models. pth can't be uploaded the ip-adapter. pickle. モデルは以下のパスに移動します。 stable-diffusion-webui\models\ControlNet Sep 11, 2023 · could share the comfyUI Ip-adapter_sdxl_face workflow or generate the image. 2. (there are also SDXL IP-Adapters that work the same way). cl ur dm qv lp yl we su cg oj